Difference between blocking and split plot-design?

Question

What is the difference between blocking in an experiment design and a split plot design?

Answer 1

What is needed to understand is what split plot design does and how it is different from blocking.

Consider a factor $A$ with $p$ levels. If an RBD is constructed with, say, $r$ blocks, $p$ levels of $A$ are allocated at random at each of the $r$ plots.

Nothing intricate.

Let there be another factor of interest $B$ with $q$ levels. How to compare with the levels of $A?$ If the former needs only a smaller amount of material as that of the latter, then one ingenious way to implement the same is to divide each whole plot into $q$ sub-plots, where $q$ levels of $B$ are allocated at random. This layout is the basis of the split plot design.

$\rm [I]$ compares SPD and RBD, the foremost of which is:

$\bullet$ Albeit there is no net gain in precision in both designs, there is an increased precision on the main effect of the sub-plot treatment $B$ and the interaction effect of $AB$ (at the cost of precision on the main effect $A$).

---

Reference:

$\rm [I]$ Experimental Designs, William G. Cochran, Gertrude M. Cox, Asia Publishing House, $1959,$ sec. $7.13,$ pp. $296-297.$